The decision this page serves. For an adult with irritable bowel syndrome (IBS), does deliberately modulating the gut microbiome — swallowing a probiotic, feeding resident microbes a prebiotic, a synbiotic combining both, or taking a gut-selective antibiotic (rifaximin) — reduce symptoms enough to be worth doing? IBS is in scope as a prevalent, modifiable-exposure-improvable condition with symptom and quality-of-life endpoints; per-patient drug selection/dosing is not, but rifaximin’s efficacy and limitations as a standard option ARE (the pharmacotherapy taper). This is the supplement/drug lever for the same condition Low-FODMAP Diet for IBS serves with the dietary lever — a companion, not a competitor.
The evidence here is one 2018 gold systematic review with meta-analysis of 66 RCTs (67 trials); it is also the kind of underlying evidence the Gut Microbiome and Health recommendation (WGO 2023) rests on — so cite Ford for effect and certainty, cite the guideline for what-to-do. The two are the same evidence base viewed at two altitudes, not independent corroboration.
Probiotics — a real class-average benefit that is strain-blind, so no strain is proven
Fifty-three probiotic RCTs (5545 patients) were pooled. Combination probiotics showed a benefit on symptom persistence: «Combination probiotics were assessed in 21 RCTs … containing 1931 patients, with a significant effect on symptoms (RR = 0.79; 95% CI 0.68 to 0.91)» (Ford et al., 2018) (NNT 7). But that pooled number is the load-bearing trap, not the finding: it carries significant heterogeneity (I2 = 72%, P < 0.001) and funnel-plot asymmetry (Egger P = 0.06, so probable publication bias), and it averages across combinations that point in opposite directions.
- The strain-heterogeneity beneath the average. Specific combinations diverge: a seven-strain mix RR 0.48 (95% CI 0.24-0.94) and LacClean Gold RR 0.59 (0.37-0.93) helped, while the L. paracasei F19 + L. acidophilus La5 + B. lactis Bb12 combination showed no benefit (RR 0.92, 0.76-1.11) and VSL#3 was non-significant (RR 0.82, 0.52-1.30). Single genera mostly failed: whole Lactobacillus RR 0.82 (0.63-1.06, no clear benefit; I2 = 83%), Saccharomyces RR 0.92 (0.82-1.03, not superior), Bifidobacterium RR 0.70 (0.48-1.01, trend only). The positive single-strain signals (L. plantarum DSM 9843 RR 0.67, 0.51-0.87, NNT 3; E. coli DSM17252; S. faecium) each rest on one to three small trials, and the L. plantarum heterogeneity persisted (I2 = 63%). (Ford et al., 2018)
- Verdict: benefit at the class-average level, but no specific product is robustly proven. Ford’s own conclusion: «Which particular combination, species, or strain of probiotics are effective for IBS remains, for the most part, unclear»; «it was not possible to draw definitive conclusions about their efficacy»; «it remains unclear which combination, species, or strain should be preferred in the individual patient.» (Ford et al., 2018) So take a probiotic for IBS is under-specified advice: the exposure is the strain, not the word.
This is exactly the discipline Gut Microbiome and Health states as WGO’s central rule — effects are
strain- and dose-specific, and strain-blind pooling «should be avoided». Ford quantifies and
demonstrates that rule inside IBS: the pooled effect exists yet dissolves the moment you ask which
organism. [inferred from @ford2018probiotics] — Ford is the granular evidence behind the guideline's discipline (type-F refinement), not an independent route to it.
Prebiotics and synbiotics — insufficient evidence, not a demonstrated null
Prebiotics: three small RCTs, no clear benefit. A 20 g fructooligosaccharide trial (n = 98) found 58.0% vs 65.2% improvement (not significant); a 2.5 g short-chain FOS trial (n = 79) was underpowered and null; a trans-galactooligosaccharide crossover (n = 60) improved global symptom scores in both dose arms versus washout but had no effect on abdominal pain. (Ford et al., 2018)
Synbiotics: two RCTs (198 patients). Individually positive but the pooled continuous estimate was non-significant because of extreme heterogeneity (SMD -1.73; 95% CI -3.73 to 0.27; I2 = 96%). (Ford et al., 2018)
Ford’s summary — «this meta-analysis has demonstrated little evidence for the use of prebiotics or synbiotics in IBS» (Ford et al., 2018) — is best read as insufficient evidence (three and two tiny trials respectively), NOT a proven no-effect. The expectancy test: with so few, small, underpowered trials, we could not yet expect to have seen the gradient even if one exists. Held as a G-gap, not a null. — adequately powered prebiotic/synbiotic RCTs would move this from insufficient toward benefit or no-effect; none is held.`
Rifaximin — the best-evidenced, but modest and non-durable, drug option
The gut-selective antibiotic rifaximin is the cleanest result in the paper. «There were five trials of similar design that used rifaximin in non-constipated IBS patients, which was more effective than placebo (RR of symptoms persisting = 0.84; 95% CI 0.79-0.90)» (Ford et al., 2018), with no heterogeneity (I2 = 0%) and NNT 9. The low-risk-of-bias subset holds: «There were four low risk of bias rifaximin trials, assessing 1966 patients … There remained a significant effect in favour of active therapy» (RR 0.87, 95% CI 0.82-0.93, NNT 11). (Ford et al., 2018) Rifaximin is now licensed for IBS-D in the US, and appeared safe (no excess adverse events; near-zero de-novo C. difficile).
The limitations that size this rock (pharmacotherapy taper — efficacy AND its limits):
- Modest and non-durable. Retreatment works but weakly: «the efficacy was modest after each course of treatment, and the long-term safety of repeated courses of rifaximin, and how many times to re-treat patients whose symptoms recur remains uncertain.» (Ford et al., 2018) Symptoms relapse; this is a repeatable short course, not a cure.
- Mechanism unclear, and not via a lasting microbiome change. «Studies that have evaluated the effect of rifaximin on the microbiome, show that any changes are limited, and are not sustained.» (Ford et al., 2018) The SIBO rationale is contested (breath tests are unreliable), so the drug works by an unresolved route.
- Cost and stewardship. A branded minimally-absorbed antibiotic taken in repeated courses carries cost and antimicrobial-resistance monitoring the dietary levers do not.
Neomycin (RR 0.73) and norfloxacin (RR 0.63) each had a single positive trial — too thin to conclude. (Ford et al., 2018)
Why honest uncertainty, not confident advice — the expectancy and bias discipline
IBS symptom endpoints are self-reported and subjective (global symptom relief, abdominal pain), so these trials sit exactly where placebo/expectancy effects are largest — the same discipline Low-FODMAP Diet for IBS is held to. Two mechanical warnings from Ford reinforce it: probiotic analyses showed funnel asymmetry (probable publication bias inflating the pooled effect), and half the probiotic trials (27/53) were only unclear risk of bias. There is also a popularity-vs-evidence gap: Ford notes «most believe probiotics to be a benign therapy» and that over 90% of surveyed clinicians use them, «despite the fact that any effect of probiotics in IBS is poorly understood» (Ford et al., 2018) — attention runs ahead of effect size, the anti-signal the telos warns of.
Conflicts of interest, appraised on the same bar. The study was funded by the American College of Gastroenterology; two authors declare industry ties relevant here — one to probiotic companies, one (advisory/speaker) to the maker of rifaximin. (Ford et al., 2018) Under symmetric standards this earns neither extra suspicion nor a pass: the review argues against easy enthusiasm (it deflates the pooled probiotic effect and calls rifaximin only modest), which is the opposite direction to the COIs, so the appraisal stands on its methods.
Decision framing (Layer 3)
- Who this is for. An adult with diagnosed IBS whose symptoms are not controlled — a real remediable gap. None of these interventions has evidence in people without IBS.
- Ranking the levers at the margin. For non-constipated / IBS-D, rifaximin is the best-evidenced single option (consistent, homogeneous, NNT 9), but modest and needing repeat courses — weigh it against the dietary lever Low-FODMAP Diet for IBS (also short-term, also subjective endpoints) as the realistic alternative, not as a first reflex. Probiotics are worth a bounded trial only if a specific studied strain/combination is chosen (e.g. L. plantarum DSM 9843, or a studied multi-strain mix), with low expectations and a stop rule — a generic probiotic is a coin-toss. Prebiotics and synbiotics are not yet supported; no recommendation either way.
- Adherence and reversibility. All are reversible and low-harm short-term; the honest framing is a time-limited trial with a planned review, never an open-ended regimen, and never a claim that these fix the underlying disorder.
Value type and standing
Provisional-C domain opener: a single gold meta-analysis induces the IBS microbial-therapy question into the fabric, as a companion cluster to Low-FODMAP Diet for IBS. It also carries a type-F refinement of Gut Microbiome and Health — quantifying the WGO strain-specificity discipline with the actual IBS pooled/subgroup numbers (same evidence base, not independent-E). Confidence medium: the rifaximin result is internally strong (I2 = 0%, low-RoB-robust) but modest; the probiotic result is a heterogeneous, publication-biased class-average with no proven strain; prebiotic/synbiotic evidence is insufficient. The banked cross-source synthesis is owed when a second independent gold SR (or a newer strain-specific network meta-analysis) lands.